meven added a comment.
In D25117#565877 <https://phabricator.kde.org/D25117#565877>, @dfaure wrote: > Nice, I didn't realize the slave had 5 seconds to cleanup after being killed, I thought it died immediately > (genericsig_handler in slavebase.cpp). I didn't realize there was a timeout somewhere. > Hmm, the Windows code is probably a no-op then, there's no signal handler there to call setKillFlags(), so no opportunity to cleanup before dying. > Some research indicates that we'd have to use `SetConsoleCtrlHandler` on Windows for similar behaviour. > https://stackoverflow.com/questions/2007516/is-there-a-posix-sigterm-alternative-on-windows-a-gentle-kill-for-console-ap > https://danielkaes.wordpress.com/2009/06/04/how-to-catch-kill-events-with-python/ I did not realize that. > OK so at least it seems feasible, so I won't object to the commit even if it doesn't implement it on Windows. This would be a regression in Windows, because of the code removal in `filecopyjob.cpp`. I need to implement signal handling for windows starting at `KIOPrivate::sendTerminateSignal` in `kioglobal_p_win.cpp`. currently it simply kills directly the process. and use `SetConsoleCtrlHandler` as you suggested to fix this. Or add a `#ifdef Q_OS_WIN` block in filecopyjob.cpp. REPOSITORY R241 KIO REVISION DETAIL https://phabricator.kde.org/D25117 To: meven, #frameworks, dfaure, ngraham, apol Cc: apol, kde-frameworks-devel, LeGast00n, GB_2, michaelh, ngraham, bruns